Optimal Redistributive Taxation Louis Kaplow Book Overview Scope: Conceptual Framework Focus on major structural features Emphasizing relationships among the pieces contrast existing Public Economics Handbook (4+ vols.) and surveys of specific subjects Not: Empirical Political economy Macroeconomics (stabilization, ...) 2 Contents I. II. III. IV. V. Framework Optimal Income Taxation Optimal Commodity Taxation Applications Tax Equity 3 Contents I. II. III. IV. V. Framework Optimal Income Taxation Optimal Commodity Taxation Applications Tax Equity 4 Framework Purposes of taxation Redistribution Revenue-raising identical individuals: trivial result that uniform (lump-sum) levy optimal; no distortion distortionary income tax (etc.) because distribution matters “equity-efficiency” tradeoff thus central Corrective (Pigouvian) 5 Integrated view (ch. 2) Optimal use of instrument A depends on whether instrument B available Notably, income tax (and transfers) are core instrument for revenue/redistribution Other tools (commodity taxes, estate/gift tax, ...) viewed as special-purpose supplements assessed on efficiency grounds – income tax adjustments can hold distribution constant Likewise for public goods, regulation, … 6 Approach Distribution-neutral adjustment to the income tax Isolates “distinctive” effects of policies (estate/gift taxation, public goods, Pigouvian taxation) Developed in detail, below, with commodity taxation 7 Social objective (ch. 3) Explicit reference to social welfare function surprisingly high use of fuzzy proxies (ability to pay) and often unspecified notions of fairness inappropriate, and often indeterminate in any event value of tracing all to welfare is large (e.g., family unit, estate/gift tax, administration/error) “Welfarism”: notion that all of social relevance can be traced to effects on individuals’ welfare some defense in chapters 13-15 see generally Fairness versus Welfare (2002) (w/ Shavell) 8 SWF SW ( x) W ui ( x) f (i )di Ex.: Utilitarianism: sum of utilities Decreasing marginal utility in u Ex: u = ln y MU = 1/y Utilitarian SWF: marginal $ worth 10 times to those with 1/10 the income 9 Common formulation ui ( x ) 1 e SW ( x ) f (i )di , for e 1 1 e ln ui ( x ) f (i )di , for e 1 e : inequality aversion parameter Also decreasing MU, in the u function Weight on redistribution depends on both But arguably mainly the latter 10 Contents I. II. III. IV. V. Framework Optimal Income Taxation Optimal Commodity Taxation Applications Tax Equity 11 Optimal taxation setup Income taxation Really a labor income tax (one-period model) Roughly equivalent to a “wage” tax or a cash-flow consumption tax in this one-period setting Commodity taxation Think of as incorporating all manner of differential taxation (including of savings) Tax expenditures (and penalties) can be analyzed as differential commodity taxation 12 Optimal income taxation (ch. 4) Problem statement: Choose tax system to max SWF, subject to: Individuals choose labor supply to max U Exogenous revenue requirement Source of equality/efficiency tradeoff: “Type” (earning ability w) assumed unobservable Tax income, a signal of ability Hence distortion because not taxing ability directly 13 Illustration of tax/transfer fn Nonlinear Income Tax and Transfer Schedule 14 Linear income tax Linear Income Tax Schedule, t = 40% and g = $12,000 15 Linear income tax: F.O.C. cov ( w), y( w) t 1 t y( w) ( w) f ( w)dw ( w) W uc ( w) l ( w) tw g Stern (1976): t=54%, g=34% of ave. inc. 16 Two-bracket income tax Slemrod et al. (1994) Two-Bracket Income Tax Schedule Upper bracket w/ lower marginal rate More so as weight on equality increases 17 Nonlinear income tax: concept Problem Perturbation thought experiment: Determine marginal tax rate at all levels of income Determine intercept: i.e., grant to individual earning zero Suppose we raise the proposed marginal tax rate on all income in the range (50,000, 50,100) by 1% (one percentage point) Two effects: Inframarginal: all individuals with y (wl) above 50,100 pay $1 more in tax Marginal: all individuals with y in the interval (50,000, 50,100) They all face same marginal tax rate as before: no substitution effect (Income effect: all work a tiny bit more) Sum: obtain significant revenue, and from higher-income individuals (where high is relative to $50,100 in this case) Pay $0 to $1 more if behavior unchanged They face higher marginal tax rate, so substitution effect induces to work less (is also an income effect) Questions: how large is each effect, what does it depend on, and how does it relate to social welfare? 18 Nonlinear income tax: F.O.C. T ( w * l*) 1 F ( w*) 1 T ( w * l*) * w * f ( w*) marginal folks (with income w*l*) inframarginal folks (income > w*l*) (1-F)/f: 1 F ( w*) Perturbation thought experiment W (u( w))uc f ( w)dw w*1 1-F is fraction of population above marginal rate in question – collect more revenue inframarginally from them as raise marginal rate at a given point f is density at margin, those who are distorted ξ*, w* Second term: average welfare cost to the group that pays the increase (for case with no income effects) Many simulations have high marginal rates that trail downward at top 19 Ability taxation (ch. 5.C) Ideal tax is ability tax Income taxed as observable signal of ability Explanation of Pareto improvement Could have full equalization Source of incentive tradeoff Ability tax surrogates possible? Height? Age? 20 Transfer payments (ch. 7) Optimal transfers: optimal income tax Categorical assistance Model as type-specific income tax (ability taxation, above) Results: phase-outs Work inducements I.e., if have figured out optimal income tax, including (plausibly negative) taxes on the poor, why aren’t we done? Hard to justify Informational assumptions Cash versus in-kind (raises different types of questions) 21 Contents I. II. III. IV. V. Framework Optimal Income Taxation Optimal Commodity Taxation Applications Tax Equity 22 Commodity taxation (ch. 6) Problem: choose relative tax rates on different commodities, given an income tax that can be adjusted Result: no differentiation optimal in basic case Atkinson & Stiglitz (1976) showed true with optimal income tax; Kaplow (2006) extends to when income tax nonoptimal Intuition: adds additional distortion without helping on the redistribution/distortion tradeoff Example: inefficiency of luxury taxes, exemptions for necessities (e.g., for food in a VAT) Assumptions and qualifications 23 Proof sketch Step 1: begin with differential commodity tax Step 2: abolish tax differentials and adjust income tax to hold utility level constant for each taxpayer (at each level of prior income) Step 3: this combined reform has no effect on labor supply intuition: utility return to every level of labor effort same (for everyone); hence labor same (see diagram) Step 4: a surplus is produced (due to reducing the distortion of consumption choices) Step 5: pro rata rebate of surplus: Pareto improvement 24 Figure 6.1 Labor Supply with Offsetting Income Tax Adjustment Summary of intuition Distribution can be ignored Labor supply effects can be ignored Because by construction is distributionneutral Because are none (qualifications coming) What’s left?: the “distinctive” efficiency consequences of the reform 26 Two-step decomposition What if a reform is not distributionneutral? Decompose as follows: (1) Construct distribution-neutral reform (2) Move from that to actual reform Note : latter is purely redistributive Comments Clarifies analysis Facilitates specialization 27 Qualifications Commodities that interact with utility of labor-leisure choice: Other qualifications higher tax rate on books, which are leisure complements, because discouraging leisure reduces labor supply distortion subsidize central city amenities Goods that signal ability (as distinct from income): opera? Externalities – e.g., pollution, gifts, … (more below) … Still, it provides general way to assimilate many fiscal issues into a common framework 28 Contents I. II. III. IV. V. Framework Optimal Income Taxation Optimal Commodity Taxation Applications Tax Equity 29 General implications Ramsey taxation principles (inverse elasticity rule) essentially wrong in this setting All analysis based thereon (e.g., of capital income taxation) not right when is an income tax that may be adjusted Examples Exemption of food from VAT (groceries vs. restaurant meals) Taxation of gifts: viewed as one sort of consumption expenditure Public goods: can compare “price” to marginal WTP Taxation of savings (income versus consumption tax, etc.): standard income tax is equivalent to pure labor income tax plus differential taxation on commodities consumed in different periods . . . 30 “Income” vs. consumption Tax and taxation of capital (ch. 9) Earn in period 1, consume in periods 1 and 2; model difference in terms of budget constraints (illustration with flat-rate tax, no grant) c2 wl (1 t ) c1 1 r c2 wl (1 t ) c1 1 r (1 t ) 31 Analysis Interpret as differential commodity taxation Consumption in different periods = consumption of two different commodities Atkinson-Stiglitz result applies Note: not pro-rich because doing it distribution-neutral (income tax adjusts) 32 Discussion Capital taxation generally Including wealth taxation Corporate income tax Recent dynamic Mirrlees literature As a form of differential capital taxation “Intertemporal wedge” Intuition: multi-period, future labor income uncertainty, induces precautionary savings, but that dampens future labor supply (wealth effect) Administrative arguments bearing on income tax vs. consumption tax 33 Corrective taxation (ch. 8.G) Pigouvian taxes and subsidies Similar features for much of regulation Qualifications? Including liability through the legal system Distributive weights Labor supply distortion (“double dividend,” etc.) Neither – again Analogy to differential commodity taxation Distribution-neutral income tax adjustment 34 Transfer (estate/gift) taxation (ch. 10) Description/model: differential commodity taxation based on own- versus other-consumption Analysis Gift externalities again, presumptively inefficient Positive, to donee Negative, income effect on donee Note: again, integrating w/ income tax (which can tax rich however would like) transforms question and produces qualitatively different analysis & results Extension: charitable giving 35 Unit of taxation (ch. 12) Marrieds vs. singles; number of children Controversial Qualitative changes over time Differences across countries Huge common denominator problem – need link to objective function (social welfare function) Approach Positive model of family (e.g., models of sharing, as in Becker 1974) SWF with each individual counting once 36 Analysis: distribution Does unequal sharing clearly favor “individualized” treatment? (no) Do scale economies clearly favor less generous treatment of marrieds? (no) Should child benefits be fixed amounts / declining with income? (no) Note: form of SWF and degree of riskaversion (rate of declining MU of consumption) have qualitative effect here 37 Analysis: incentives Work Marriage Child-bearing 38 Contents I. II. III. IV. V. Framework Optimal Income Taxation Optimal Commodity Taxation Applications Tax Equity 39 Framework (ch. 13 & 14) Welfarism Restate definition Other norms can lead to Pareto conflicts Proxy rationales Choice of social welfare function “Original position” Sufficiently egalitarian? Whose welfare included? 40 Other tax equity norms (ch. 15) E.g., Ability to pay Horizontal equity Benefit principle Sacrifice principle Definitions as norms (e.g., use of Haig-Simons definition of “income”) Criticism Need root in social welfare Vague, incoherent Conflict with Pareto principle Possible value as proxy principles 41